Kyros Engineering Engineering Tools
Home › Tools › Filter Designer Pro
// Analog tools Proto v0.1ButterworthChebyshevBesselSallen-Key

Filter Designer Pro

Pick a response, an order and a cutoff. You get the Bode, group-delay and step plots, and a parts list for each op-amp stage, rounded to E96 and E24 with the error that rounding costs. A digital FIR tab covers windowed and Parks-McClellan designs.

Runs entirely in your browser. Nothing is uploaded.

1 · Domain

Digital IIR (Butterworth, Chebyshev, elliptic biquads, CMSIS-DSP export) lives in the IIR Biquad Designer.

2 · Response

Hz
Hz
dB

Butterworth and Bessel edges are −3 dB points. Chebyshev edges are where the response leaves the ripple band.

3 · Circuit

F
×

Band-pass response uses MFB band-pass stages. Band-stop lists each notch section's f0 and Q.

4 · Order helper

Smallest order meeting a spec for the chosen approximation (low-pass: pass edge below stop edge; high-pass: the reverse).

dB
dB
Hz
Hz
—

5 · Plot range

Hz
Hz

2 · FIR design

Hz
Hz
Hz
Hz
×
Hz
Hz

Result

Magnitude

Phase

Group delay

Step response

Stages and component values

Poles and zeros

In rad/s, left half-plane. Zeros on the imaginary axis for band-stop.


      

Coefficients


      
// How it works

From a pole list to resistor values

An analog filter is a set of poles (and zeros) in the s-plane. A prototype low-pass at 1 rad/s gives the shape. Frequency transforms then move it to the cutoff you asked for, or turn it into a high-pass, band-pass or band-stop. Every plot is evaluated from the poles directly, so the magnitude, phase and group delay are exact rather than numerically differentiated.

H(s) = K · Π(s − z_i) / Π(s − p_i) low-pass s → s/ωc high-pass s → ωc/s band-pass s → (s² + ω0²)/(BW·s) band-stop s → BW·s/(s² + ω0²) ω0 = √(ωL·ωH) BW = ωH − ωL group delay τ(ω) = Σ Re[1/(jω − p_i)] − Σ Re[1/(jω − z_i)]

The three approximations

Stages

Poles are grouped into conjugate pairs. Each pair is a second-order section with its own natural frequency and Q, and each real pole is a first-order RC. The tool lists stages lowest-Q first, which keeps the high-Q stage from clipping early. The circuits below use an ideal op-amp. Pin names match the component table.

Sallen-Key LP Vin–R1–A, A–R2–B, B–C2–GND, A–C1–Vout (B = follower input) ω0 = 1/√(R1R2C1C2) Q = √(R1R2C1C2) / (C2(R1+R2)) R1 = R2 = 1/(2Qω0C2), C1 = 4Q²C2 Sallen-Key HP Vin–C1–A, A–C2–B, A–R1–Vout, B–R2–GND Q = ω0·C1C2R2/(C1+C2) R1 = 1/(2Qω0C), R2 = 2Q/(ω0C) MFB LP Vin–R1–A, A–C1–GND, A–R3–B(−), A–R2–Vout, B–C2–Vout gain −K = −R2/R1 C1 ≥ 4Q²(K+1)C2 R2 = (1 − √(1 − 4Q²(K+1)C2/C1)) / (2ω0QC2) R3 = 1/(ω0²R2C1C2) MFB HP Vin–C1–A, A–R1–GND, A–C3–B(−), A–C2–Vout, B–R2–Vout gain −C1/C2 = −K C1 = C3 = C, C2 = C/K R2 = (2K+1)Q/(ω0C) R1 = K/(ω0·C·Q·(2K+1)) MFB BP Vin–R1–A, A–R2–GND, A–C1–B(−), A–C2–Vout, B–R3–Vout centre gain −G C1 = C2 = C R1 = Q/(Gω0C) R2 = Q/((2Q²−G)ω0C) R3 = 2Q/(ω0C)

E96 and E24 columns round each part to the nearest standard value by ratio. The tool then recomputes f0 and Q from the rounded parts, so you see what the rounding costs. Tolerances of the parts are on top of that. At Q above about 3, use 1 % parts and expect to trim.

FIR designs

A windowed-sinc design multiplies the ideal impulse response by a window. The cutoff lands at −6 dB, and the transition width is about 3.3/N of the sample rate for Hamming. Parks-McClellan places the error evenly instead: you give a transition width and a stopband weight, and the Remez exchange finds the taps with the smallest worst-case ripple. The same tap count then gives deeper stopband attenuation.

windowed: h[n] = w[n] · sin(π·fn·(n−M)) / (π·(n−M)), fn = fc / (fs/2), M = (N−1)/2 Parks-McClellan: minimise max |W(ω)·(D(ω) − A(ω))| over the pass and stop bands group delay of a symmetric FIR = (N−1)/2 samples at every frequency

Assumptions and limits

References

  1. A. B. Williams, F. J. Taylor, Electronic Filter Design Handbook, 4th ed., McGraw-Hill, 2006: normalized pole tables for Butterworth, Chebyshev and Bessel filters, step overshoot figures.
  2. J. Karki, Active Low-Pass Filter Design, Texas Instruments application report SLOA049B, and Analysis of the Sallen-Key Architecture, SLOA024B.
  3. Analog Devices, tutorial MT-206 and the other filter tutorials in the MT-2xx series: normalized filter tables and stage design.
  4. A. I. Zverev, Handbook of Filter Synthesis, Wiley, 1967: band-pass and band-stop prototypes.
  5. W. E. Thomson, "Delay networks having maximally flat frequency characteristics," Proc. IEE 96(44), 1949: Bessel filters.
  6. J. H. McClellan, T. W. Parks, L. R. Rabiner, "A computer program for designing optimum FIR linear phase digital filters," IEEE Trans. Audio Electroacoustics 21(6), 1973. The tool's exchange code was written from this description.
  7. A. V. Oppenheim, R. W. Schafer, Discrete-Time Signal Processing, 3rd ed., Pearson, 2010: windowed FIR design, Kaiser window.
  8. SciPy, scipy.signal: butter, cheby1, bessel, firwin, remez, buttord (BSD-3-Clause). The tool's tests compare against scipy 1.18; no scipy code is used at run time.

Filter on paper, noise on the bench?

Kyros designs complete analog front ends: sensor interface, anti-alias filtering, converter and the firmware behind it, with the layout and tolerance work that makes the numbers hold in production.

Talk about your project →